[Swift-devel] Strange errors while staging in data

[David Kelly]

I made a small change to CacheNode.java in my local copy that just avoids
calling any methods on FutureObject f if it is null. Maybe there is a
better fix, but it seems to be working well for now anyway.

Index:
modules/karajan/src/org/globus/cog/karajan/compiled/nodes/CacheNode.java
===================================================================
---
modules/karajan/src/org/globus/cog/karajan/compiled/nodes/CacheNode.java
(revision 3874)
+++
modules/karajan/src/org/globus/cog/karajan/compiled/nodes/CacheNode.java
(working copy)
@@ -128,7 +128,9 @@

  Cache cache = getCache(stack, key == this);
  synchronized (cache) {
  FutureObject f = (FutureObject) cache.getCachedValue(key);
- f.setValue(ret);
+ if(f != null) {
+   f.setValue(ret);
+ }


On Sat, Mar 1, 2014 at 1:35 PM, David Kelly <davidkelly at uchicago.edu> wrote:

> Hello,
>
> At the beginning of my run (apparently before any jobs actually start on
> the scheduler) I am running into this error:
>
> Swift 0.95 RC5 swift-r7605 cog-r3874
>
> RunID: 20140301-1911-fe86tqf8
> Progress: Sat, 01 Mar 2014 19:11:49+0000
> Progress: Sat, 01 Mar 2014 19:11:50+0000  Selecting site:193  Stage in:307
>
> Execution failed:
> Exception in RunpSIMS:
>     Arguments: [047, 438, params.psims, output/047/438output.tar.gz]
>     Host: midway
>     Directory: RunpSIMS-20140301-1911-fe86tqf8/jobs/g/RunpSIMS-g4b3m5nl
> exception @ swift-int.k, line: 530
> Caused by: java.lang.NullPointerException
> cache @ swift-int.k, line: 134
> Caused by: java.lang.NullPointerException
> at
> org.globus.cog.karajan.compiled.nodes.CacheNode.setValue(CacheNode.java:131)
> at
> org.globus.cog.karajan.compiled.nodes.CacheNode.runBody(CacheNode.java:77)
> at
> org.globus.cog.karajan.compiled.nodes.InternalFunction.run(InternalFunction.java:153)
>
>
> This only seems to happen when I have 500 or more tasks. If I set
> foreach.max.threads to a low number like 50, it seems to work fine.
>
> The log is at
> http://web.ci.uchicago.edu/~davidk/logs/RunpSIMS-20140301-1921-r4dps0i4.log
> .
>
> Any ideas?
> Thanks,
> David
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/swift-devel/attachments/20140302/3e740857/attachment.html>